Use when defending the research design of a Communication Research (CR) manuscript — experiments (lab/online), panel surveys, and content analysis with intercoder reliability. CR is quantitative and demanding about identification, measurement, and confound control. Strengthens the design; it does not write code.
NewMaintained
Automatic
◆Research pack
Last verified 9/2/2026
What it changes
Passed 20 of 25 test cases with the skill, and 20 of 25 without it.
Without skill80%
With skill80%
Checks it passed that the model alone missed
Requires explicitly specifying the unit of analysis
1 fixed19 passed either way4 still failing1 broke
Pass-rate lift
0 pp
Tokens
+47%
Agent turns
0%
Response time
-6%
Model gemini-3.6-flashMeasured 9/2/2026v120 of 25 cases passed with skill
Tokens, turns and time describe resource use, not a dollar price.
See it in action
A test case the skill fixed. Same model, same prompt, with and without the skill.
Prompt
For a quantitative content analysis of political news coverage, two research assistants coded 5,000 news articles. When coders disagreed, they discussed discrepancies until 100% consensus was reached. The authors report 100% final consensus reliability. Evaluate this reliability reporting approach for a Communication Research submission and specify the exact reporting and sampling requirements for intercoder reliability.
What was checked, without and with the skill
Checked
Without skill
With skill
Rejects consensus reached after discussion as an intercoder reliability metric
Passed
Passed
Requires reporting Krippendorff's alpha on an independent subsample
Passed
Passed
Requires explicitly specifying the unit of analysis
Missed
Passed
case-06 · gemini-3.6-flash · 9/2/2026 · answers are capped and redacted
What's inside
The instructions your model reads when the skill loads.
SKILL.md107 linessingle file
Research Design (commres-research-design)
CR is a quantitative journal and demanding about each design. The design must credibly connect the
hypotheses (commres-theory-building) to evidence and defeat the strongest rival explanation. This
skill is mode-aware — pick the section that matches your study and defend it on social-science terms.
When to trigger
Specifying an experiment, panel survey, or content-analysis protocol
A reviewer questioned causal claims, sampling, coding reliability, validity, or a confound
Preparing a preregistration / pre-analysis plan (note it in the cover letter)
Justifying why your design adjudicates the rival account from commres-literature-positioning
Experiments (lab / online / survey-embedded)
Preregister design and primary analyses; report a-priori power / MDE; pre-specify subgroups.
Stimulus sampling: treat messages as a sample, not a fixture — multiple exemplars per condition;
model message as a random factor so the feature effect is not one text's idiosyncrasy.
Manipulation and attention checks; treatment realism; report and model attrition.
Pre-specify the mediation/moderation test that operationalizes H2/H3; for causal mediation,
prefer manipulating the mediator or a measurement-of-mediation design with stated assumptions.
Ethics/IRB and informed consent; debrief where deception is used.
Surveys / panels
Sampling frame, mode, and generalization claims; weighting where appropriate.
Validated multi-item measures; report reliability (alpha/omega) and, ideally, a CFA/measurement
model; guard against common-method variance (procedural and statistical remedies).
For process claims, prefer panel (lagged) or experimental leverage — cross-sectional mediation
cannot license a temporal/causal story; say so plainly if you are cross-sectional.
Content analysis
A documented codebook; trained coders; report intercoder reliability (Krippendorff's alpha
or equivalent) on an adequate subsample, and the unit of analysis.
Justify text sampling (timeframe, sources); establish construct validity of categories, not just
reliability — reliable coding of the wrong construct is still wrong.
Use when defending the research design of a Comparative Political Studies (CPS) manuscript — cross-national/panel identification, case-based comparison and process tracing, experiments, or multi-method designs. CPS prizes comparative leverage. Strengthens the design; it does not write code.